Density-Functional Calculations of the Adsorption and Reaction of

Apr 3, 2008 - Pedro O. Bedolla , Gregor Feldbauer , Michael Wolloch , Christoph Gruber , Stefan J. Eder , Nicole Dörr , Peter Mohn , Josef Redinger ,...
0 downloads 0 Views 216KB Size
J. Phys. Chem. C 2008, 112, 6947-6952

6947

Density-Functional Calculations of the Adsorption and Reaction of Acetic Acid on Ge(001) Hyung-Jin Kim and Jun-Hyung Cho* BK21 Program DiVision of AdVanced Research and Education in Physics, Hanyang UniVersity, 17 Haengdang-Dong, Seongdong-Ku, Seoul 133-791, Korea ReceiVed: February 26, 2008

The adsorption and reaction of acetic acid on the Ge(001) surface are investigated by first-principles densityfunctional calculations. We find that the carbonyl O atom initially bonds to the down Ge atom of the buckled dimer, and subsequently, O-H dissociation takes place through the intradimer or interdimer transfer of the H atom. The resulting monodentate (MD) structure after the intradimer H-atom transfer remains stable, while that after the interdimer H-atom transfer proceeds to formation of the end-bridged (EB) bidentate structure across the ends of two adjacent dimers by forming an additional Ge-O bond on the opposite side of the H-atom transfer. It is also possible that the latter MD structure proceeds to formation of the on-top (OT) bidentate structure on a single Ge dimer, but it is kinetically difficult because of the existence of a repulsive interaction between an O lone pair and a single Ge dangling bond. However, our calculated energy profiles for the reaction pathways predict that, as temperature increases, the relatively less stabilized MD and EB structures are converted into the most stable OT structure, as observed by a recent scanning tunneling microscopy experiment.

1. Introduction The hybrid organic-silicon systems provide a favorable linking of organic chemistry to existing silicon-based microelectronic technologies, and therefore the useful properties of organic molecules such as light emission or light detection may be tailored to silicon-based devices.1,2 Over the past decade, the reactions of simple and conjugated alkene molecules with the Si(001) surface have been extensively studied, resulting in substantial development in understanding the reaction mechanism of the CdC functional group on Si(001).3-14 Recently, such efforts to understand hybrid organic-silicon systems have been extended to other classes of organic compounds containing nitrogen and oxygen, and further expanded such hybridization with the Ge(001) surface.15-22 Since the carboxyl group is an important functional group of DNA and proteins, the detailed knowledge about the reactions of carboxylic acids with semiconductor surfaces would provide useful information related to the reactions of biomolecules with semiconductor surfaces. There have been several experimental studies15-20 for the adsorption of various carboxylic acids on the Si(001) surface, showing that carboxylic acids undergo O-H dissociation. Recently, the additional surface reactions of carboxylic acids producing interdimer products have been reported on the Ge(001) surface.21,22 Using Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, and DFT calculations, Filler et al.21 elucidated that carboxylic acids on Ge(001) undergo O-H dissociation at 310 K. They also suggested the presence of a bidentate bridging structure in which both oxygen atoms interact directly with the surface. Subsequently, the STM experiment of Hwang et al.22 not only confirmed the bidentate structure which was proposed by Filler et al.,21 but also observed three different adsorption configurations which were assigned to the monodentate (MD) structure and the two kinds of bidentate structures such as the end-bridged * [email protected].

(EB) and on-top (OT) bidentate structures. Here, the MD structure has a single Ge-O bond between the carbonyl O atom and the down Ge atom of a buckled dimer (see Figure 1a), the EB structure has an additional Ge-O bond between the carbonyl O atom of the dissociated hydroxyl group and the Ge atom of a neighboring dimer (see Figure 1c), and the OT structure has two Ge-O bonds on the same dimer (see Figure 1e). According to the STM data for the adsorption of acetic acid on Ge(001),22 acetic acid occupies either the MD or the EB structure at room temperature, but, after annealing at 400 K, the MD and EB structures are converted into the OT structure. There have been a number of theoretical and experimental studies for the intradimer and interdimer reactions of organic molecules on the Si(001) surface.23-27 Especially, for N-H dissociation and O-H dissociation, both the intradimer and interdimer H-atom transfers were reported to be equally probable.25-27 However, we have to notice that all previous21,22 DFT calculations for the reaction of acetic acid on Ge(001) considered only the intradimer H-atom transfer, and are therefore unable to provide a proper interpretation for the three adsorption configurations observed by the STM experiment.22 In this paper, using first-principles density-functional calculations within the generalized-gradient approximation, we investigate not only the binding energy and structure of adsorbed acetic acid on Ge(001), but also the reaction pathways for acetic acid adsorption. We identify the three different reaction pathways that result in formations of the observed MD, EB, and OT configurations. Hereafter, the three reaction pathways are labeled as I, II, and III, respectively. The initial reaction of acetic acid with Ge(001) occurs via a precursor in which the electron-deficient down Ge atom of the buckled dimer attracts a lone pair of the carbonyl O atom to form a Ge-O bond. Subsequently, along the reaction pathway I (II and III), O-H dissociation of the hydroxyl group easily takes place through the intradimer(interdimer) transfer of the H atom, leading to the MDI (MDII) structure as shown in Figure 1a (1b). The MDI

10.1021/jp801666y CCC: $40.75 © 2008 American Chemical Society Published on Web 04/03/2008

6948 J. Phys. Chem. C, Vol. 112, No. 17, 2008

Kim and Cho TABLE 1: Calculated Adsorption Energies (in eV) of Acetic Acid on the Ge(001) Surface, in Comparison with Previous Theoretical Resultsa path I II III

clusterb LDAc this this this

P

TP-MD

MD

0.72

0.58

0.63 0.63 0.63

0.48 0.35 0.35

1.71 1.70 1.23 1.01 1.01

TMD-BD

BD

1.42 (0.57) 0.77 0.56 0.33

1.61 2.10 (2.39) 1.17 1.08 1.39

a P, T, MD, and BD denote the precursor, transition, monodentate, and bidentate structures, respectively. All data are taken from the solid lines displayed in Figures 2, 3, and 4. TP-MD (TMD-BD) represents the transition state from P to MD (from MD to BD). The BD structures along the reaction pathways I, II, and III correspond to the EBI, EBII, and OT structures, respectively. The values in parentheses represent the results regarding the OT structure which is formed by diffusion of the dissociated H atom. b From ref 21. c From ref 22.

Figure 1. Optimized geometries of adsorbed acetic acid on Ge(100): (a) the monodentate structure MDI, (b) the monodentate structure MDII, (c) the end-bridged bidentate structure EBI, (d) the end-bridged bidentate structure EBII, and (e) the on-top bidentate structure OT. The circles represent Ge, C, O, and H atoms with decreasing size.

structure can proceed to a bidentate structure (designated as the EBI structure in Figure 1c) by forming the second Ge-O bond with a neighboring dimer. However, the MDI structure is likely to be more occupied than the EBI structure because the former is thermodynamically more stable than the latter. On the other hand, along the reaction pathways II and III, the MDII structure is less stable than the EBII (see Figure 1d) and OT (see Figure 1e) bidentate structures. Here, we find that the two reaction pathways from the MDII structure to the EBII and OT structures show a drastic difference in adsorption kinetics. Along the reaction pathway II, formation of the EBII structure is kinetically facilitated with a low activation barrier, while along the reaction pathway III, formation of the OT structure is kinetically prohibited with a relatively higher activation barrier. These disparate features along the reaction pathways II and III originate from the different behaviors in the electronic interactions during the second Ge-O bond formation, i.e., the attractive interaction between an O lone pair state and an empty Ge dangling-bond state, and the repulsive interaction between an O lone pair state and a single-occupied Ge dangling-bond state, respectively. We note that the OT structure is much more thermodynamically stable than the MDI and EBII structures. Thus, our calculated energy profiles predict that, as temperature increases, the reverse reactions along the pathways I and II as well as the reaction along the pathway III are thermally activated, thereby attaining the most stable OT structure. This provides an explanation for the STM data,22 where after annealing the OT configuration appeared by the conversion of the MD and EB configurations. 2. Methods The total energy and force calculations were performed by using first-principles density-functional theory28,29 within the generalized-gradient approximation (GGA). We used the exchange-correlation functional of Perdew et al.30 for the GGA. The norm-conserving pseudopotentials of Ge and H atoms were constructed by the scheme of Troullier and Martins.31,32 For C and O atoms whose 2s and 2p valence orbitals are strongly localized, we used the Vanderbilt ultrasoft pseudopotentials.33 The surface was modeled by a periodic slab geometry. Each slab contains six Ge atomic layers, and the bottom Ge layer is passivated by two H atoms per Ge atom. The thickness of the

vacuum region between these slabs is about 15 Å, and acetic acid molecules are adsorbed on the unpassivated side of the slab. To make the interaction between adsorbed molecules sufficiently weak, we employed a 2 × 4 unit cell that involved four dimers along the dimer row. The electronic wave functions were expanded in a plane-wave basis set using a cutoff of 25 Ry, and the electron density was obtained from the wave functions at two k points in the surface Brillouin zone of the 2 × 4 unit cell. All the atoms except the bottom two Ge layers were allowed to relax along the calculated Hellmann-Feynman forces until all the residual force components were less than 1 mRy/bohr. Our calculation scheme has been successfully applied for the adsorption and reaction of various unsaturated hydrocarbon molecules on Ge(001).34 3. Results We first determined the atomic structure of adsorbed acetic acid on Ge(001) within the MD, EB, and OT configurations. Each optimized structure is shown in Figure 1. The calculated adsorption energies (Eads) for these structures are given in Table 1, together with those obtained by previous cluster21 and local density approximation (LDA)22 calculations. There are two kinds of MD structures, MDI (Figure 1a) and MDII (Figure 1b), which are formed through the intradimer and interdimer transfer of the H atom, respectively. We find that the MDI (MDII) structure has an adsorption energy of 1.23 (1.01) eV. Each MD structure can proceed to formation of the bidentate structure with two Ge-O bonds. We considered three possible bidentate structures, i.e., the EBI (Figure 1c), EBII (Figure 1d), and OT (Figure 1e) structures. Here, the EBI structure is formed from the MDI structure along the reaction pathway I, whereas the EBII and OT structures are formed from the MDII structure along the reaction pathways II and III, respectively. We find that the EBI, EBII, and OT structures have Eads ) 1.17, 1.08, and 1.39 eV, respectively. Thus, the stability of various adsorption structures is ordered as Eads(OT) > Eads(MDI) > Eads(EBI) > Eads(EBII) > Eads(MDII).35 This order is somewhat different from the previous LDA13 result of Eads(OT) ) 2.39 eV > Eads(EBI) ) 2.10 eV > Eads(MDI) ) 1.70 eV, that is, the relative stability of the MDI and EBI structures is reversed between the present GGA and previous LDA calculations. We note that the present stability of the MDI and EBI structures with ∆Eads ) 0.06 eV agrees well with the previous cluster calculation of Filler et al.21 where the MDI structure is more stable than the EBI structure by ∆Eads ) 0.10 eV. The above adsorption energies obtained from the previous LDA22 calculations are considerably larger than those of the present GGA calculations because of the wellknown overestimation36 of the adsorption energy in the LDA.

Adsorption/Reaction of Acetic Acid on Ge(001)

J. Phys. Chem. C, Vol. 112, No. 17, 2008 6949

Figure 2. Calculated energy profile for the reaction pathway I, forming the MDI and EBI structures. The atomic geometries of several representative points are given: the precursor state PC, the transition state TC-I, the MDI structure, the transition state TI, and the EBI structure. One additional intermediate state between TI and EBI is also displayed. The dashed line represents the energy file for O-H dissociation through the precursor state PH. Energy is referenced from the gas state. Dark circle represents the Ge atom participating in the second Ge-O bond formation.

A recent STM experiment for adsorbed acetic acid on Ge(001) observed three different adsorption configurations such as the MD, EB, and OT configurations.22 Interestingly, the MD and EB configurations were observed at room temperature, while the OT configuration appeared only after annealing at 400 K. This result indicates that at room temperature acetic acid is trapped at either the MD or the EB configuration, but, as temperature increases, both configurations are transformed to the OT configuration. We note that the previous LDA calculation of Hwang et al.22 considered only the intradimer H-atom transfer at a precursor where the carbonyl O atom weakly bonds to the down Ge atom of the buckled dimer, thereby leading them to assign the three observed configurations to the MDI, EBI, and OT structures. To explain the above STM22 data, we study the three reaction pathways I, II, and III, which result in formation of the three observed adsorption configurations. The Ge(001) surface consists of rows of buckled dimers, where a partial charge transfer occurs from the down to the up Ge atom,37 therefore being partially positively and negatively charged, respectively. The electron-deficient down Ge atom attracts the lone pair of acetic acid, leading to the formation of a Ge-O bond where the lone pair state of the O atom hybridizes with the empty danglingbond state of the Ge atom. Since acetic acid has the two kinds of O atoms (one is carbonyl O in the CdO moiety and the other is hydroxyl O in the OsH moiety), there are two possible precursor states (designated as PC and PH: see Figure 2). In the PC (PH) state, the carbonyl (hydroxyl) O atom bonds to the down Ge atom of the buckled dimer. We find that the PC state has Eads ) 0.63 eV, larger than that (Eads ) 0.12 eV) of the PH state. This indicates that the bonding of the carbonyl O atom to the down Ge atom is relatively stronger than that of the hydroxyl O atom. Similar to the reaction of other carboxylic acids on the Si(001) surface,15-20 the above-mentioned precursor states of adsorbed acetic acid on Ge(001) proceed to dissociation of the O-H bond. It is noticeable that there are two possible ways for the dissociation. One is the intradimer transfer of the H atom along the reaction pathway I, and the other is the interdimer

transfer of the H atom along the reaction pathways II and III. First, we focus on the intradimer transfer of the H atom at the PC state. To examine the cleavage of the O-H bond, we calculated the energy profile for the dissociation by optimizing the structure with increasing the bond length dO-H. Here, we optimized the structure by using the gradient projection method38 where only the distance (but not angles) between the O atom and the dissociating H atom was constrained. HellmannFeynman forces are used for the relaxation of all the atomic positions as well as the O-H bond angles for each of several values of the O-H bond length. In this way, we find a pathway from the PC state to the MDI structure along which the energy is a minimum everywhere for fixed dO-H. The atomic geometry of the transition state (designated as TC-I) on going from the PC state to the MDI structure is displayed in Figure 2. We find that the TC-I state has Eads ) 0.48 eV which is 0.15 eV smaller than that of the PC state, yielding an energy barrier (Eb) of 0.15 eV from PC to MDI. We also calculated the energy profile for O-H dissociation at the PH state (see the dashed line in Figure 2). We obtain Eb ) 0.59 eV from PH to MDI, much larger than that (Eb ) 0.15 eV) from PC to MDI. These values of Eb are in good agreement with the previous cluster21 calculation where Eb for the O-H dissociation through the PC (PH) state was calculated to be 0.14 (0.57) eV. Note that the adsorption energy (Eads ) 0.12 eV) of the PH state is smaller than the corresponding O-H dissociation barrier. Thus, it is likely that desorption from the PH state would be more easily activated rather than O-H dissociation. On the basis of the calculated energy profiles for O-H dissociation, we can say that formation of the MDI structure via the PC state is kinetically and thermodynamically favored over that via the PH state. Along the reaction pathway I, the MDI structure can proceed to formation of the EBI structure where the carbonyl O atom of the dissociated hydroxyl group bonds to the Ge atom of a neighboring dimer. We have to note that, during this Ge-O bond formation, there exists a strong repulsive interaction between the O lone pair electrons and the dangling-bond electrons of the electron-abundant up Ge atom (represented by a dark circle in MDI; see Figure 2) of a neighboring dimer.

6950 J. Phys. Chem. C, Vol. 112, No. 17, 2008

Kim and Cho

Figure 3. Calculated energy profile for the reaction pathway II, forming the MDII and EBII structures. The atomic geometries of several representative points are given: the precursor state PC, the transition state TC-II, the MDII structure, the transition state TII, and the EBII structure. The dashed line represents the energy file for O-H dissociation through the precursor state PH.

However, we find that, if this Ge atom is transformed from the up-buckled (nucleophilic) side to the down-buckled (electrophilic) one via the flip-flop39,40 motion of the Ge dimer at room temperature, formation of the Ge-O bond takes place with no barrier. This process corresponds to a section of the reaction pathway I ranging from the MDI to the EBI structure in Figure 2. Here, we obtained Eb ) 0.46 eV for the buckling switch of the neighboring dimer. Note that, in the TI state, the structure of the neighboring dimer is symmetric. Using an Arrheniustype activation process with the usual attempt frequency of ∼1013 Hz for the preexponential factor, the flipping rate of the buckled dimer at room temperature is estimated to be ∼2 × 105 s-1, thereby easily attaining the EBI structure. However, since the MDI structure is more stable than the EBI structure by ∆Eads ) 0.06 eV, we estimate that at room temperature the MDI structure is ∼10 times more likely occupied than the EBI structure. Next, we study the reaction pathway II from the PC state to the EBII structure. The calculated energy profile along the reaction pathway II and the atomic geometries of various intermediate and transition states are displayed in Figure 3. We find that the interdimer H-atom transfer at the PC state easily occurs with Eb ) 0.28 eV, forming the MDII structure. On the other hand, the energy barrier for the transition from the PH state to the MDII structure is Eb ) 0.57 eV. Thus, similar to the case of intradimer H-atom transfer, the interdimer H-atom transfer occurs nearly through the PC state rather than through the PH state. After formation of the MDII structure, the carbonyl O atom of the dissociated hydroxyl group can be attracted to the down Ge atom of the neighboring dimer, leading to formation of the EBII structure. This process for the second Ge-O bond formation is analogous to that occurring from MDI to EBI along the reaction pathway I. We find that the TII state in which the structure of the neighboring dimer is symmetric has Eads ) 0.56 eV, yielding Eb ) 0.45 eV from the MDII to the EBII structure. Here, unlike along the reaction pathway I where MDI is more stable than EBI by ∆Eads ) 0.06 eV, EBII is more stable than MDII by ∆Eads ) 0.07 eV (see Table 1). Thus, at room temperature the EBII structure is estimated to be ∼15 times more likely occupied than the MDII structure.

Our calculated energy profiles for the reaction pathways I and II lead us to believe that the MD and EB configurations observed by the STM experiment of Hwang et al.22 should be assigned mostly to the MDI and EBII structures. This assignment to the observed MD and EB configurations differs from the previous LDA22 calculations where the MDI and EBI structures were assigned. We note in the previous LDA calculations that the EBI structure is more stable than the MDI structure by ∆Eads ) 0.40 eV. This energetics for the MDI and EBI structures cannot explain the observed coexistence of the MD and EB configurations at room temperature. According to the STM22 data, the population of the EB configuration is larger than that of the MD configuration by 50%. This greater population of the EB configuration may be accounted for if we consider that there are two directions (due to the presence of two neighboring dimers) available to the interdimer H-atom transfer producing the EBII structure while one to the intradimer H-atom transfer producing the MDI structure. Finally, we study the reaction pathway III which results in formation of the OT structure. The calculated energy profile and the atomic geometries of intermediate and transition states along the reaction pathway III are displayed in Figure 4. Since the MDII structure is an intermediate state before formation of the OT structure, the reaction pathways II and III have an identical energy profile up to the MDII structure. In order to obtain the energy barrier from the MDII to the OT structure, we rotate the molecule (from the MDII structure) about the axis of the Ge-O bond to form the second Ge-O bond. At each of several rotation angles, we relaxed all the atoms to their equilibrium positions, enforcing the constraint of rotation angle by keeping the positions of two O atoms (including the pivotal Ge-O bond) in the same plane. We find Eads ) 0.33 eV for the TIII structure (whose optimized structure is shown in Figure 4), yielding Eb ) 0.68 eV from the MDII to the OT structure. This energy barrier is larger than that (Eb ) 0.45 eV) from MDII to EBII. On the basis of these calculated energy barriers, we can say that the transition from MDII to EBII is kinetically favored over that from MDII to OT by a factor of ∼103 at room temperature. We note that, as temperature increases, the reverse reaction from EBII to MDII will be kinetically feasible because

Adsorption/Reaction of Acetic Acid on Ge(001)

J. Phys. Chem. C, Vol. 112, No. 17, 2008 6951

Figure 4. Calculated energy profile for the reaction pathway III, forming the OT structure. The atomic geometries of several representative points are given: the precursor state PC, the transition state TC-II, the MDII structure, the transition state TIII, and the OT structure.

its activation barrier is only 0.52 eV. As a result, it is likely that the EBII structure is dominantly occupied at room temperature but will be gradually converted to the OT structure with increasing temperature. Unlike the reverse reaction from EBII to MDII, the reverse reaction from OT to MDII is kinetically difficult because of the existence of a high activation barrier (Eb ) 1.06 eV). These results provide an explanation for the STM22 data where the EB configuration is formed at room temperature but after annealing at 400 K is converted into the OT configuration. To explain the STM22 observation of the conversion from the MD to the OT configuration, we have to consider formation of the OT structure through a reaction pathway traversing the MDI and MDII structures. Since the transition from MDI to MDII occurs via the Pc state, its activation barrier can be estimated to be 0.75 eV from the reverse barrier from MDI to Pc. For this reverse reaction, the Arrhenius analysis gives a rate of ∼4 × 103 s-1 at 400 K, indicating that the transition from MDI to MDII would be slower than that (with a rate of ∼3 × 106 s-1) from EBII to MDII. Thus, we expect in future STM experiments that, as temperature increases, the conversion from EBII to OT will be observed before that from MDI to OT. We note that Hwang et al.22 considered the possibility that the OT structure could be transformed from the EBI and MDI structures by diffusion of the dissociated H atom. However, such a possibility is unlikely because of the existence of a high activation barrier for the H-atom diffusion on Ge(001).41 It is noteworthy that there are disparate features for the second Ge-O bond formation along the reaction pathways II and III: that is, there is no barrier after the buckling switch of a neighboring dimer along the reaction pathway II (see Figure 3), while there is a high-energy barrier of 0.68 eV along the reaction pathway III (see Figure 4). This difference in kinetics can be accounted for by the different electrostatic interactions of the two orbitals (one is the O lone pair state and the other is the Ge dangling-bond state) participating in the Ge-O bond formation. Along the reaction pathway II (III), there is an attractive (repulsive) interaction between the lone pair state of the carbonyl O atom of the dissociated hydroxyl group and the empty (single-occupied) dangling-bond state of the reacting Ge atom represented by dark circle in Figure 3 (4).

4. Summary We have performed first-principles density-functional calculations for the adsorption of acetic acid on the Ge(001) surface. Our calculations not only identify the three different reaction pathways that result in formations of the observed MD, EB, and OT configurations, but also provide an assignment of the observed MD and EB configurations with the MDI and EBII structures, respectively. The calculated energy profiles for the three reaction pathways predict that, as temperature increases, the relatively less stabilized MDI and EBII structures are converted into the most stable OT structure, consistent with the STM measurement. The present results for various reaction pathways and adsorption structures of acetic acid on Ge(001) can be generally applicable to the adsorption of other carboxylic acids on Ge(001). Acknowledgment. This work was supported by grants from the Basic Research Program (No. R01-2006-000-10920-0) of the Korea Science & Engineering Foundation and the MOST/ KOSEF through the Quantum Photonic Science Research Center. References and Notes (1) Wolkow, R. A. Annu. ReV. Phys. Chem. 1999, 50, 413. (2) Bent, S. F. Surf. Sci. 2002, 500, 879, and references therein. (3) Nishijima, M.; Yoshinobu, J.; Tsuda, H.; Onchi, M. Surf. Sci. 1987, 192, 383. (4) Yoshinobu, J.; Tsuda, H.; Nishijima, M.; Onchi, M. J. Chem. Phys. 1987, 87, 7332. (5) Liu, H.; Hamers, R. J. J. Am. Chem. Soc. 1997, 119, 7593. (6) Hamers, R. J.; Hovis, J. S.; Lee, S.; Liu, H.; Shan, J. J. Phys. Chem. 1997, 101, 1489. (7) Hamaguchi, K.; Machida, S.; Nagao, M.; Yasui, F.; Mukai, K.; Yamashita, Y.; Yoshinobu, J.; Kato, H. S.; Okuyama, H.; Kawai, M.; Sato, T.; Iwatsuki, M. J. Phys. Chem. 2001, 105, 3718. (8) Hovis, J. S.; Hamers, R. J. J. Phys. Chem. 1997, 101, 9581. (9) Liu, Q;, Hoffmann, R. J. J. Am. Chem. Soc. 1995, 117, 4082. (10) Konecˇny, R.; Doren, D. J. Surf. Sci. 1998, 169, 417. (11) Choi, C. H.; Gordon, M. S. J. Am. Chem. Soc. 1999, 121, 11311. (12) Sorescu, D. C.; Jordan, K. D. J. Phys. Chem. 2000, 104, 8259. (13) Cho, J.-H.; Kleinman, L. Phys. ReV. B 2001, 64, 235420; Cho, J.H.; Kleinman, L. Phys. ReV. B 2004, 69, 075303. (14) Fan, X. L.; Zhang, Y. F.; Lau, W. M.; Liu, Z. F. Phys. ReV. B 2005, 72, 165305. (15) Ikeura-Sekiguchi, H.; Sekiguchi, T. Surf. Sci. 1999, 433-435, 549. (16) Lu, X.; Zhang, Q.; Lin, M. C. Phys. Chem. Chem. Phys. 2001, 3, 2156.

6952 J. Phys. Chem. C, Vol. 112, No. 17, 2008 (17) Bitzer, T.; Alkunshalie, T.; Richardson, N. V. Surf. Sci. 1996, 368, 202. (18) Bitzer, T.; Richardson, N. V. Surf. Sci. 1999, 427-428, 369. (19) Lopez, A.; Bitzer, T.; Heller, T.; Richardson, N. V. Surf. Sci. 2001, 480, 65. (20) Hwang, H.-N.; Baik, J.-Y.; An, K.-S.; Lee, S.-S.; Kim, Y.-S.; Hwang, C.-C.; Kim, B.-S. J. Phys. Chem. B 2004, 108, 8379. (21) Filler, M. A.; Van, Deventer, J, A.; Keung, A. J.; Bent, S. F. J. Am. Chem. Soc. 2006, 128, 770. (22) Hwang, E.; Kim, D. H.; Hwang, Y. J.; Kim, A.; Hong, S.; Kim, S. J. Phys. Chem. C 2007, 111 (16), 5941; Kim, D. H.; Hwang, E.; Hong, S.; Kim, S. Surf. Sci. 2006, 600, 3629. (23) Lu, X.; Zhu, M. Chem. Phys. Lett. 2004, 393, 124. (24) Minary, P.; Tuckerman, M. E. J. Am. Chem. Soc. 2005, 127, 1110. (25) Qu, Y. Q.; Wang, Y.; Li, J.; Han, K. L. Surf. Sci. 2004, 569, 12. (26) Chung, O. N.; Kim, H.; Chung, S.; Koo, Phys. ReV. B 2006, 73, 033303. (27) Lee, J. Y.; Cho, J.-H. J. Phys. Chem. B 2006, 110, 18455. (28) Hohenberg, P.; Kohn, W. Phys. ReV. 1964, 136, B864. (29) Kohn, W.; Sham, L. J. Phys. ReV. 1965, 140, A1133. (30) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77, 3865. (31) Troullier, N.; Martins, J. L. Phys. ReV. B 1991, 43, 1993. (32) Kleinman, L.; Bylander, D. M. Phys. ReV. Lett. 1982, 48, 1425. (33) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892; Laasonen, K.; Pasquarello, A.; Car, R.; Lee, C.; Vanderbilt, D. Phys. ReV. B 1993, 47, 10142. (34) Cho, J.-H.; Kim, K.; Morikawa, Y. J. Chem. Phys. 2006, 124, 024716; Kim, H. J.; Cho, J.-H. J. Chem. Phys. 2004, 120, 8222; Cho, J.H.; Kleinman, L. Phys. ReV. B 2003, 67, 115314.

Kim and Cho (35) We have also performed spin-polarized DFT calculations for various adsorption geometries such as MDI, MDII, EBI, EBII, and OT structures. However, all structures are found to be non-spin-polarized. (36) Sprik, M.; Hutter, J.; Parrinello, M. J. Chem. Phys. 1996, 105, 1142, and references therein. (37) Chadi, D. J. Phys. ReV. Lett. 1979, 43, 43. (38) Wismer, D. A.; Chattery, R. Introduction to Nonlinear Optimization; North-Holland: Amsterdam, 1978; pp 174-178. (39) As for the ground state reconstruction of the Si(001) and Ge(001) surfaces, density functional calculations on slab geometries and large cluster models favor a buckled dimer structure, see: Dabrowski, J.; Scheffler, M. Appl. Surf. Sci. 1992, 56, 15; Ramstad, A.; Brocks, G.; Kelly, P. J. Phys. ReV. B 1995, 51, 14504; Konecny, R.; Doren, D. J. J. Phys. Chem. B 1997, 101, 10983; Yang, C.; Lee, S. Y.; Kang, H. C. J. Chem. Phys. 1997, 107, 3295. On the other hand, multiconfiguration self-consistent field and configuration interaction calculations on small clusters find the symmetric dimer structure to have the lowest energy: see Radeke, M. R.; Carter, E. A. Phys. ReV. B 1996, 54, 11803; Paulus, B. Surf. Sci. 1998, 408, 195; Shoemarker, J.; Burggraf, L.; Gordon, M. S.; W. J. Chem. Phys. 2000, 112, 2994; Gordon, M. S.; Shoemarker, J.; Burggraf, L. W. J. Chem. Phys. 2000, 113, 9355. (40) Fukada, Y.; Shigeta, Y. Phys. ReV. Lett. 2003, 91, 126103. (41) In their LDA calculations, Hwang et al. (ref 13) obtained Eb ) 1.13 eV for the diffusion barrier of the dissociated H atom. Using temperature-programmed desorption, Russell and Ekerdt (see Surf. Sci. 1996, 369, 51) estimated that the H diffusion barrier on Ge-covered Si(001) surface is 1.0 eV.